Modelling rainfall in selected agricultural hubs in Nigeria: a comparative probability distributions study

The choice of the most suitable probability distribution for rainfall varies across different geographical regions due to varying spatial and temporal variability. Therefore, there is an ongoing need to conduct research aimed at determining the optimal distribution models for various locations. This study was specifically designed to investigate the variability in rainfall patterns and identify the most appropriate probability distribution models for total monthly rainfall in selected agricultural areas in Nigeria. To achieve this, a comprehensive analysis was conducted on 41 years rainfall data spanning from 1982 to 2022. Ten different probability distributions were applied to fit these data for each calendar month as well as for the overall dataset. Applicable parameters (shape, scale, location, rate, logarithmic mean, and logarithmic standard deviation) for each distribution were estimated using maximum likelihood estimates. The selection of the best-fit distribution model for each month was based on three criteria, including the Akaike information criteria (AIC), Bayesian information criteria (BIC), and corrected Akaike information criteria (CAIC). The results revealed that the Pearson III and loglogistic distributions provided the best-fit model for rainfall data during most months at selected locations. However, when considering the overall rainfall across most locations, the Gamma distribution exhibited superior performance compared to the other distributions. This research contributes to the enhancement of integrated water resources management and facilitates the accurate evaluation and planning of water resources in selected agricultural hubs in Nigeria, with the ultimate aim of improving agricultural productivity.